Key takeaways
- Phishing and spoofing was the most reported cybercrime to the FBI in 2025, with 191,561 complaints.
- ABA Opinion 477R says lawyers must establish policies and periodically train staff in reasonably secure electronic communication.
- Train by role: reception, bookkeeping, lawyers and remote staff face different attacks.
- Cover AI-enabled scams such as cloned voices and polished fake emails; the FBI received more than 22,000 AI-related complaints in 2025.
- Measure how many people report a simulated phish, not only how many click.
Why training belongs in the security budget
Most attacks on law firms start with a person, not a server: an email that looks like it came from a court, a client asking to change bank details, a call from 'IT' asking for a sign-in code. The FBI's Internet Crime Complaint Center received 191,561 phishing and spoofing complaints in 2025, more than any other category, along with 24,768 business email compromise complaints[1].
Training is also an ethics expectation. ABA Formal Opinion 477R lists training lawyers and nonlawyer assistants in technology and information security among its considerations, and says that lawyers must establish policies and procedures and periodically train employees, subordinates and others assisting in the delivery of legal services in the use of reasonably secure methods of electronic communication[2]. It ties this to the supervisory duties in Rules 5.1 and 5.3[2].
Training is also inexpensive compared with the alternatives. An hour of staff time each quarter and a simulation tool cost far less than a single diverted settlement payment or a week without email, and the same habits reduce the odds of both.
What staff need to be able to do
Good training is about behaviors, not awareness for its own sake. By the end of the first year, every person at the firm should be able to:
- Spot the common signs of a phishing message and report it with one click or one call.
- Refuse to change payment details without a call-back to a known number.
- Recognize a sign-in prompt they did not start, and deny and report it.
- Send sensitive documents only through the approved secure channel.
- Handle a lost device or a suspicious click by reporting it immediately, without fear of blame.
- Know what not to put into public AI tools.
A 12-month plan
Month 1: baseline
Run a 45-minute session for everyone, including partners, on phishing, payment fraud, MFA prompts and reporting. Send a first simulated phishing email a week later to set a baseline.
Months 2โ3: payment fraud
Walk the bookkeeper, paralegals and lawyers who handle money through the call-back procedure using a realistic settlement scenario.
Months 4โ5: devices and remote work
Cover lost phones, home offices, public Wi-Fi and smart speakers, with the firm's own policies as the material.
Month 6: tabletop
Run a one-hour incident exercise with the incident response team and a few staff volunteers.
Months 7โ9: AI-enabled scams
Show examples of convincing fake emails and explain voice-cloning requests, plus the firm's rules for using AI tools with client information.
Months 10โ11: confidentiality habits
Phone call privacy, voicemail, messaging channels and what to do when a client overshares at intake.
Month 12: review
Compare simulation results with the baseline, gather feedback and set next year's topics.
Send one simulated phishing email each month throughout the year, varying the theme. Keep each training session short and specific to the firm. Generic slide decks about hackers in hoodies do not change behavior.
Designing simulations that teach
| Simulation | What it tests | Example measure |
|---|---|---|
| Fake court e-filing notice with a link | Checking sender and link before clicking | Example: share who reported it within an hour |
| Client asking to update wiring details | Applying the call-back rule | Example: share who escalated instead of replying |
| Shared-document request from an unknown sender | Credential phishing awareness | Example: share who entered credentials on the fake page |
| Urgent request from a 'partner' on a new email address | Authority pressure | Example: share who verified by phone |
| Unexpected MFA prompt (announced exercise) | Denying and reporting unknown prompts | Example: number of reports to the office manager |
Make simulations realistic but fair. Use themes staff actually see, never threats about pay, layoffs or personal emergencies, and follow every simulation with a short explanation of the clues. The goal is recognition, not humiliation.
Training by role
- Reception and intake: callers impersonating clients or IT, requests to confirm whether someone is a client, and what not to say on voicemail.
- Bookkeeping and anyone handling trust funds: payment instruction changes, fake invoices and vendor impersonation.
- Lawyers: targeted emails using real case names, document-sharing lures, and the temptation to bypass rules when in a hurry.
- Remote and hybrid staff: home networks, shared devices, smart speakers and working in public.
- Partners and managers: they approve exceptions and set the tone, so they need to see that the rules apply to them first.
Keep role modules short, around fifteen minutes, and use the firm's own documents, systems and near misses as examples. A bookkeeper learns more from walking through the firm's actual wire approval screen than from any generic video, and lawyers pay attention when the fake email in the exercise uses a real court's name and format.
AI-enabled scams and AI use
AI has made old scams more convincing. The FBI reports receiving more than 22,000 complaints with AI-related information in 2025, with adjusted losses above $893 million, and notes that chat generators can quickly create official-sounding emails while voice cloning can be used to request wire payments; businesses reported losses over $30 million to business email compromise involving AI[1]. Staff should know that a familiar voice on the phone is no longer proof of identity, and that the call-back rule applies even to a partner's voice.
The other half is how staff use AI tools themselves. Thomson Reuters found that 64% of professionals had received no training on using generative AI for their industry work, even though 89% saw use cases for it[3]. Cover which tools are approved, what client information may never be entered into public tools, and how to check AI output before relying on it.
Making reporting easy and blame-free
Training only works if reporting is easier than ignoring a problem. Build the reporting route before the first session:
- One button or one address for suspicious emails, and one phone number for anything urgent, printed on a card at every desk.
- A promise, stated by a partner, that nobody is punished for reporting a mistake promptly, including their own click.
- A quick reply to every report, even if it is only 'thanks, this one was safe', so people know reports are read.
- A short monthly note to staff about what was reported and what the firm learned, without names.
- Recognition for good catches, especially from people outside the lawyer ranks.
People hide mistakes when they expect blame, and attackers count on that delay. The firms that recover fastest from real incidents are usually the ones where the first person to notice a problem picked up the phone within minutes.
Measure the right things
Click rates are easy to track and easy to game. The more useful number is how many people report a suspicious message, and how quickly. A firm where most people report within the hour will catch a real attack early even if someone clicks. Track reporting rates, time to report, and repeat clicks by the same person, and share firm-wide results without naming individuals.
Visibility is part of the problem training solves. In the ABA's 2023 survey, 19% of respondents did not know whether their firm had ever experienced a security breach[4]. Staff who report freely give the firm a far better picture of what is actually happening.
Example: from blame to reporting
Example (fictional): a twelve-person firm's first simulation finds that several people clicked and only one reported. Instead of naming the clickers, the managing partner thanks the person who reported at the next staff meeting and installs a 'report phish' button in the email client. Over the following months, reports rise steadily. In month eight, a paralegal reports a real email impersonating a client within ten minutes; the office manager blocks the sender before anyone else responds. The partner shares that story, anonymized, at the next session.
Related guides
Training works best alongside clear policies. See data retention and destruction policies, HIPAA and medical records in injury firms for records handling, the threat data in law firm cyberattack statistics and trends, and safe AI use in protecting client data when using AI tools. New intake staff can fold security into their 30-day onboarding plan.
Frequently asked questions
Are law firms required to provide security training?
ABA Opinion 477R says lawyers must establish policies and periodically train staff in reasonably secure electronic communication, grounded in the supervision duties of Rules 5.1 and 5.3.
How often should law firms run phishing simulations?
Monthly simulations with short follow-up explanations work well for most firms, alongside a few brief training sessions during the year.
Should staff be disciplined for clicking a simulated phish?
Generally no. Blame discourages reporting. Focus on coaching and on raising the reporting rate; reserve escalation for repeated disregard of clear rules.
What should AI training for law firm staff cover?
Which tools are approved, what client information must never be entered into public tools, how to verify AI output, and how AI makes scams such as voice-cloned payment requests more convincing.
References
Federal Bureau of Investigation, Internet Crime Complaint Center (2026). 2025 IC3 Annual Report (opens in a new tab) โ
American Bar Association Standing Committee on Ethics and Professional Responsibility (2017). Formal Opinion 477R: Securing Communication of Protected Client Information (archived copy) (opens in a new tab) โ
Thomson Reuters (2025). Generative AI Adoption Nearly Doubles as Professional Services Reach Crossroads (opens in a new tab) โ
American Bar Association, Law Practice Division (2023). 2023 Cybersecurity TechReport (archived copy) (opens in a new tab) โ